sporthoj.py

import requests
from bs4 import BeautifulSoup
import pickle
from urllib.request import urlopen
import json
from urllib.parse import quote_plus
import sys
import os

# Letar efter nya annonser på Sporthoj.com

template = """Ny annons på Sporthoj.com.
Titel: {title}
Pris: {price}
Länk: https://sporthoj.com/annonser/id/{id}

{description}"""
storage = os.path.join(sys.path[0], 'id')

failures = 0

try:
    next_id = pickle.load(open(storage, 'rb'))
except:
    next_id = 10458

while True:
    next_id += 1
    print('Trying to load ad #{0}'.format(next_id))
    page = requests.get("https://sporthoj.com/annonser/id/{0}".format(next_id))
    if (failures == 10):
        # This probably means that there are no new ads
        break
    if (page.status_code != 200):
        failures += 1
        continue

    failures = 0
    soup = BeautifulSoup(page.content, 'html.parser')
    parsed_title = soup.select('h1')[0].get_text()
    parsed_price = soup.find_all('span', class_ = 'buysell-price')[0].get_text()
    parsed_description = soup.select('p')[0].get_text()
    message = template.format(
        id = next_id,
        title = parsed_title,
        price = parsed_price,
        description = parsed_description)
    #print(message)
    libsignal = urlopen("https://signal.stormhub.org/v1/send?recipient=46733907485&message={0}&apikey=bastian-8a85xM5e5Rbfak".format(quote_plus(message)))
    response = json.loads(libsignal.read())
    print('-> ' + json.dumps(response, indent = 4, sort_keys = True))
    pickle.dump(next_id, open(storage, 'wb'))